Apply rotation to one or a list of data.

Determine how many samples are random data / streaming data.

Sample `num` items from a `lst` by taking every k-th element.

Swish - Described in: https://arxiv.org/abs/1710.05941

Swish - Described in: https://arxiv.org/abs/1710.05941

Convert to BBOX_DTYPE np.array.

In place conversion to torch (from numpy).

Similar to transpose this list of lists. Assume we have [Label([None, None, bbox, None]), Label([None, None, bbox, None]), Label([bbox, None,

r"""Fills the input Tensor with values drawn from a truncated normal distribution. The values are effectively drawn from the normal distributi

Apply NMS on predicted bboxes.

Apply NMS on predicted bboxes. Input: preds: List[(N_i, 7)], [(xyxy), obj_conf, cls_conf, cls_idx] Returns: output: same as

Update the lens by adding.

Verify whether the newly generated data is correct.

Takes a bbox in the form [x1,y1,x2,y2] and returns a new bbox in the form [x,y,w,h] where x,y is the center and w,h are the width and heigh

Apply zoom in to one or a list of data.

Apply zoom out to one or a list of data.

Add two ObjectLabelBase objects by concating.

List concatenation.

:param data_dict: LoaderDataDictGenX type, image-based tensors must have (*, h, w) shape. :return: map with same keys but spatially a

Apply weak and strong augmentation to get two versions of output.

Check if two ObjectLabelBase objects are equal.

Load a frame with its bbox labels, and `L` event reprs before it.

Load a sub-seq of event data + labels.

Load the bbox labels for the item-th frame.

Initialises a tracker using initial bounding box.

Sets key parameters for the tracker.

Sets key parameters for the tracker.

:param desired_hw: Desired height and width :param mode: See torch.nn.functional.pad :param value: See torch.nn.functional.p

ctor :param datfile: binary dat or npy file

In case of fastmode == True: use uint8 to construct the representation, but could lead to overflow. In case of fastmode == False: use